labour market
Slovenia is a country with a relatively high degree of automation and robotisation39 in comparison to other Central and Eastern European countries.40 The share of enterprises using robots41 reaches 7% in Slovenia, which is equal to the EU average. Within that, 6% of robots are industrial and 1% service robots (5%
and 2% respectively in the EU as a whole (Eurostat)). In 2018 Slovenia was ahead of other EU-CEE11 countries in the share of enterprises using robots in the food, textile, wood-processing, paper and chemical industries.
Bulgaria ranked first in metals manufacturing, Slovakia in machinery production and Poland in construction (Reiter 2019). The gaps in robotisation levels between the EU-CEE11 and the EU15 are nevertheless still wide, at over 20 pps in some industries (food, textiles, wood, paper, chemicals and metals). In machinery the differences are narrower (11 pps), primarily owing to the closer integration of Eastern European countries with Western Europe in car production. Automation, however, is a broader notion than robotisation and therefore more difficult to measure.
Central and Eastern European countries, including Slovenia, have a low degree of automation compared to older EU Member States but are making very rapid progress in this area. The International Federation of Robots (IFR) estimates that the number of industrial robots in the Central and Eastern European countries will record 22% yearly growth in the next three years, compared to only 5% in Germany (IFR 2018). This is a consequence of several factors. Being “followers” of technological development trends, these countries have a lot of potential for automating simpler business and industrial processes. Despite rapid growth, their automation levels are still only at 17% of that in other EU countries (Reiter 2019). These countries also have significantly lower wages on average, although labour shortages, due to the favourable business cycle and strong demographic pressure, have increasingly affected
39 Automation means using computer programs and other technology to carry out tasks that would otherwise be done by human workers. These tasks can be mechanical or virtual, from the very simple, routine and repetitive to the highly complex. Robotics, on the other hand, involves the use of robots, which consist of computer systems, sensors, motors and other components and are programmed to perform physical tasks (for example heavy lifting) autonomously. The two have a lot in common, but many types of automation have nothing to do with physical robots, while many types of robots have nothing to do with automation.
40 i.e. the 11 countries that joined the EU after 2004.
41 The share of enterprises using robots is Eurostat’s indicator, which, together with robot density, is used to evaluate the degree of automation in a particular country or sector. In addition to industrial robots, which perform routine and clearly structured tasks, the indicator also covers service robots, which have a certain degree of autonomy and can perform intended tasks in interaction with people.
affecting the quality of life and, consequently, health is access to housing. Slovenia has a higher share of foreign citizens overburdened by housing costs and living in overcrowded dwellings than the EU on average. It thus has a significantly higher share of foreign citizens living in overcrowded dwellings than the EU average and – at the same time – the widest gap between the shares of foreign and domestic citizens living in overcrowded dwellings.
Stronger co-operation between different institutions and the adjustment of integration measures to the needs of different groups of migrants would facilitate their integration into society. For greater efficiency of measures, it is necessary to strengthen the cooperation between institutions at the local level (schools, employment services, health institutions, non-governmental organisations, etc.), which play an important role in integrating and empowering immigrants to become active members of society, as highlighted by the OECD in its publication “Working Together for Local integration of Migrants and Refugees”
(2018). Reinforcing and adjusting integration measures to different migrant groups (economic migrants, family members, vulnerable groups, etc.) with different needs would also increase their possibilities for successful integration into society. .
cognitive skills and therefore more educated workforce, which necessitates reallocation of workforce between tasks, workplaces or even sectors. This in turn increases the needs for lifelong learning. At the same time, these jobs also generate higher value added and enable an economy to improve its productivity and move up the global value chain. Automation of physically demanding tasks with low value added can also allow a society to reallocate workforce to occupations that are more appreciated in that society, or necessary, such as occupations in health and social work (long-term care), which are in greater and greater demand because of population ageing.
In mitigating the negative social effects of automation, the preparedness of economies to change plays an important role. Reallocation of workers between jobs and sectors cannot happen immediately or without disrupting the labour market, as the automation risk is unevenly distributed among workers. Those with lower or upper secondary education and qualifications are at higher risk, which has tangible social consequences for more vulnerable members of the society. Digital transformation can be successful only if accompanied by effective systems of primary and lifelong learning, re-qualifications, flexicurity and social protection that can help workers find or hold the jobs in the rapidly changing world.
The extent of job destruction and technological unemployment that will be caused by automation and robotisation is very unclear. Due to technological advancements in robotics, robots are increasingly able to perform not only routine but also non-routine manual tasks, where humans still have a comparative advantage.
The tasks that robots cannot do are persistently shrinking.
Moreover, it is increasingly clear that, besides destroying jobs, automation is transforming the nature of existing jobs, requiring workforce to adjust. Assessments of how automation is likely to affect jobs differ significantly.
Besides from the choice of data sources and methods used to categorise tasks, most differences between findings arise from the approach taken in the analysis, with some authors focusing on occupations (an occupation-based approach) and others on tasks (a task-based approach). Recently the task-based approach has become regarded as more credible in the literature, given that each occupation is comprised of a set of different tasks, of which some could be codified, while others could not.
With an ever greater prevalence of robots and other technologies in the work process, automation is going to change the demand for certain skills and jobs. Thus far all technological changes have increased workforce demand over the long term, even if in other, higher technology segments (ESDE 2018). The actual impact of automation on technological unemployment will thus depend not only on technological progress, wage growth in recent years. Further automation is one
of possible responses to these challenges and may be the best way to ensure long-term productivity and value added growth in the context of population ageing.
Automation and robotisation are increasingly changing the labour market and jobs. If 15 years ago robots were deemed capable only of performing repetitive, routine and non-cognitive tasks, the number of tasks they cannot yet do is persistently shrinking.
The OECD estimates42 that 14% of all jobs in OECD countries are at risk of automation due to technological progress and an additional 32% are likely to be significantly transformed due to the introduction of new technologies.43 With 25% of all existing jobs threatened by automation, Slovenia ranks among the countries with high job automation risk.
Automation can mitigate labour shortages in the short term. This is especially important in view of ever stronger demographic effects and relatively rapid economic growth. In the context of labour shortages, ever faster wage growth and low interest rates, it is becoming more and more sensible for companies to automate production or part of the work process. This may lead to greater prevalence of robots in companies.
Globally, three million industrial robots are expected to be in operation in 2020, double the figure in 2014.44 The studies estimate that on average one robot can substitute for around six workers in production.45 The speed at which new technologies will spread is difficult to assess, as it depends on several factors, but it is likely that part of the workforce will be replaced by robots in the short to medium term, particularly in activities that involve a lot of routine, not yet highly automated tasks. These include jobs in trade and manufacturing, i.e.
sectors where Slovenia has the most job vacancies.
Over the long term, however, automation may in fact increase demand for labour, but the newly created jobs will require more skills and higher education. All technological changes in history displaced certain jobs in the short term but led to increased demand for labour at the aggregate level in the long term. Each new technology creates occupations that may not have existed before. At the same time, it changes old occupations and, thanks to new capabilities and adaptations, facilitates access to work for less represented groups, such as disabled and older people. Newly created jobs however require more
42 Nedelkoska and Quintini (2018)
43 The estimates of the automation risk are based on expert opinion on which types of tasks are already technologically substitutable.
However, they do not take into account other factors that are equally important for the actual speed of automation, such a firms’ decisions regarding investment in automation, which also depend on prices of robots versus labour, interest rates, the investment cycle and social preferences with regard to automation.
44 International Federation of Robotics (2018), “Robots double worldwide by 2020”.
45 Acemoglu in Restrepo (2017).
Box 5: An overview of methodologies assessing the automation risk
Using the occupation-based approach, Frey and Osborne (2013), in one of the first studies of this kind, estimated the risk of automation for 702 jobs on the basis of expert opinions on technological capability. They assessed that 47% of all occupations in the US were at risk of automation.1 Their model is based on the assumption that, besides all routine tasks, it is also possible to automate all non-routine tasks that are not identified as engineering bottlenecks, i.e. tasks that we are not yet able to codify. These are tasks related to perception and manipulation, particularly where they are performed in unstructured situations, tasks related to creative intelligence, such as coming up with original ideas, and tasks that necessitate social intelligence, such as understanding other people’s reactions in social contexts and caring for others. Lordan (2018), using Autor and Dorn (2013) definitions of a
“routine task intensity” for each occupation, similarly assess that 37% to 69% of current jobs are highly susceptible to automation.
Authors supporting the task-based approach caution that the occupation-based approach may overestimate the automation risks, given that even occupations with the highest probability of automation involve a number of tasks that resist automation. Applying the task-based approach, which results in much lower estimates of automation risk, Arntz, Gregory and Ziehran (2016), using PIAAC data from 2012, estimated the automatability of jobs for 21 OECD countries, breaking down jobs into different kinds of tasks. They classified them into routine versus non-routine tasks, manual versus abstract/cognitive contents, and more versus less interactive. Tasks that could be automated include repetitive routine tasks that involve physical labour and cognitive tasks that require the collection and processing of information. Considering the heterogeneity of tasks within occupations, the authors find than only a few jobs are in fact automatable, from 6% to 12% in different countries.
The most frequently cited newer OECD study (2018), using a similar methodology, finds that 14% of jobs in 32 OECD countries are at high risk of automation,2 while another 32% could be radically transformed because of the automation of individual tasks. Using PIAAC data, the study builds on the Frey and Osborne (2013) definition of engineering bottlenecks, applying it to more narrowly defined occupational groups to get more accurate results.
These studies, however, assess the theoretical probability of automation only with regard to the technological potential and the nature of tasks. They do not assess the actual speed at which new technologies will be adopted or the likelihood of their implementation, which could be affected by several factors, from regulations on workers’
dismissal, unit labour costs, the investment cycle and decisions of individual firms to social norms and preferences with regard to automation.3 The estimates of automation risks should therefore be treated with caution. More reliable than assessments of automation risk for individual countries are relative comparisons between countries.
The assessments of automation risk for occupations differ significantly across countries. Cross-country differences reflect, among other things, differences in the sectoral composition of the economies. Sectors that are most likely to be automated are those that are still based on routine tasks, such as low-skilled jobs in manufacturing and certain craft and office-clerical jobs. According to the OECD (2018), the risk of automation is highest (33% of all jobs) in the Slovak Republic and lowest in Norway (6%). Jobs in Nordic and Anglo-Saxon countries and the Netherlands are less vulnerable to automation than those in Eastern and Southern European countries and Germany. Slovenia is among the countries with higher automation risks4 according to this analysis (25%).
An important finding of these studies is that more and more jobs require competences that cannot be automated.
The tasks that are difficult to automate (the so-called bottlenecks to automation) and are increasingly sought after require analytical and social skills, especially a combination of both.5 On the other hand, workers in fully automatable jobs are more likely to become unemployed, work fewer hours and have lower hourly wages than those in jobs with lower risk (Nedelkoska & Quintini, 2018). Outcomes of these studies indicate that automation already has a strong impact on the labour market.
1 Their study contributed to the debate on technological unemployment due to persistently high unemployment rates in advanced economies, which some analysts attributed to the increasing prevalence of computer-controlled equipment.
2 A high risk of automation means at least 70% probability of automation.
3 For example, even though, technically speaking, robots could replace carers in nursing homes, it is still more desirable for social care to be delivered by people. This is one of the reasons why – although individual tasks are performed with relative ease – the social care sector is not under pressure of automation and is increasingly hiring people to meet the rising demand.
4 A high risk of automation means that more than 50% of a worker’s tasks are likely to be displaced by machines.
5 Jobs that require a combination of skills that are not usually associated with the same job (referred to as “hybrid jobs” in the media) have become increasingly common in recent years. (https://www.weforum.org/agenda/2019/03/are-you-ready-for-the-rise-of-hybrid-jobs/).